How best to define axial spondyloarthritis (axSpA) flares in practice remains the subject of some debate as evidenced by the discussion that followed an abstract presentation at the 12th International Congress on Spondyloarthritides.
It’s an important topic, said Maxime Breban, MD, PhD, of Ambroise Paré Hospital in Paris, as flares can adversely affect patient outcomes. The absence of flares may also a useful measure of how well a patient is responding to treatment in clinical trials and whether a treatment can be tapered.
“There have been many ways to define flares in the past and there is no consensus,” he observed.
Although the Assessment of Spondyloarthritis International Society (ASAS) devised 12 preliminary definitions of flare in 2016, “these were not that good when we moved to patients,” Breban suggested.
The ASAS definitions were based on patient vignettes, he explained, and used a combination of variables from the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), a visual analog scale (VAS) of pain, and the Ankylosing Spondylitis Disease Activity Score based on C-reactive protein (ASDAS-CRP).
The study that Breban presented looked at the performance of the ASAS preliminary definitions of axSpA flares in a real-life patient population, as well as prospectively determining how variations in BASDAI and VAS pain were associated with patient-perceived flares of disease.
A total of 99 patients took part in the study, recruited through a secure e-health platform called SPONDY+. Once a week, patients completed the BASDAI questionnaire and the pain VAS, and stated whether their disease had flared in the past week.
Receiver operating characteristic (ROC) curves were calculated to see how well the BASDAI and pain VAS identified patients who were experiencing a flare or had a recently resolved flare of axSpA.
Breban reported that variation in the BASDAI “appears a suitable variable to monitor the occurrence and resolution of patient-reported flare in axial spondylarthritis.”
In predicting a flare, the area under the curve (AUC) was significantly higher for the change in BASDAI than for the change in pain VAS, at a respective 0.81 and 0.77 (P = .01). However, both variables were similarly accurate in predicting the resolution of a flare, with respective AUCs of 0.78 and 0.80 (P = .3).
Dr Désirée van der Heijde
A 0.22-point increase in BASDAI was reported to be the best balance between sensitivity (70%) and specificity (79%) for a flare. However, this is “outside of what is possible within a test–retest situation,” Désirée van der Heijde, MD, PhD, of Leiden (the Netherlands) University Medical Center, said during discussion.
van der Heijde told Breban: “I understand that that comes out of your data, that that’s the best combination for sensitivity and specificity, but the next step is to decide if that makes sense.”
The ROC curves that Breban presented showed the range of sensitivities and specificities that could be achieved. If the specificity was increased to be 90% or higher, the specificity fell to 55%, with the change in BASDAI being an increase of 0.8 points. Conversely, bringing the sensitivity above 90% meant the specificity dropped to 39% and the change in BASDAI was a decrease of 0.1 point.
“So that means you can choose whatever you want as a cutoff,” Breban said. It depends on what you are aiming to do. “If you want to identify a flare, you can increase sensitivity, or specificity, according to what your purpose is,” he suggested.
“The next step, of course, is what to choose as a flare. Then it depends on how you want to use a flare if you want to use a flare to change the treatment,” agreed van der Heijde. “That was why, in the ASAS group, it was decided to have a high specificity so that you are not changing treatment all the time.”
In the data that Breban presented, the ASAS preliminary definitions were highly specific but lacked sensitivity. None of the ASAS definitions yielded sensitivity values higher than 37%, whereas specificity was higher than 95% for all of them.
The study’s design did not allow researchers to test the ASDAS-CRP as a definition of flare in its real-world patient sample. Thus, it is looking only at the patient’s perspective on flare, and there is a “huge discrepancy” between patient and physician-reported disease activity, van der Heijde noted. “So, I think before using your data to really choose the flare definition, I think we need to take it all into account.”
Dr Maxime Dougados
Maxime Dougados, MD, PhD, of Cochin Hospital in Paris, who has been “deeply involved in the elaboration of the definition of flare” added his thoughts: “Flare means for me, not a status, but a change,” he observed.
But if the aim of treating people with axSpA is to achieve a good or acceptable state of health, he questioned whether work should be continued to define the concept of a flare.
The definition of a flare was conceived for use in clinical trials mainly, van der Heijde noted. It helped to assess how changes in treatment might affect the outcomes of patients. In clinical practice, especially now with treat-to-target gaining more and more traction in axSpA, she agreed that perhaps the goal should be to focus more on the health status of patients.
Breban acknowledged that the SPONDY+ platform has been developed by bepatient with support from Merck Sharp & Dohme. No other disclosures were made.
This article originally appeared on MDedge.com, part of the Medscape Professional Network.